From 4ecbf23189986743fad238e9692e35b1777074bb Mon Sep 17 00:00:00 2001
From: Maximilian Schmidt <maximilian.schmidt@ims.uni-stuttgart.de>
Date: Tue, 16 Jan 2024 13:52:35 +0100
Subject: [PATCH] Add DE
---
main.py | 74 +++++++++++++++++++++++++++++++++++++++++++++++----------
1 file changed, 61 insertions(+), 13 deletions(-)
diff --git a/main.py b/main.py
index afa23ec..4344fbe 100644
--- a/main.py
+++ b/main.py
@@ -146,9 +146,8 @@ Prompt 3: {prompt3}
Basic Prompt: {basic_prompt}
"""
-
-@log_calls("Performing prompt evolution")
-def evolution(prompt1: str, prompt2: str):
+@log_calls("Performing prompt evolution using GA")
+def evolution_ga(prompt1: str, prompt2: str):
# Following the evolutionary operators in GA, a new candidate prompt is generated through
# a two-step process based on the selected two parents:
# 1) The parent prompts undergo crossover, resulting in a new prompt that
@@ -172,6 +171,29 @@ def evolution(prompt1: str, prompt2: str):
return evolved_prompt
+@log_calls("Performing prompt evolution using DE")
+def evolution_de(prompt1: str, prompt2: str, basic_prompt: str, best_prompt: str):
+ # TODO add comment from paper
+ evolved_prompt = llm(
+ messages=[
+ {
+ "role": "user",
+ "content": DE_PROMPT.format(
+ prompt1=prompt1,
+ prompt2=prompt2,
+ prompt3=best_prompt,
+ basic_prompt=basic_prompt,
+ ),
+ }
+ ],
+ max_tokens=None,
+ stop="</prompt>",
+ )
+ if "<prompt>" in evolved_prompt:
+ evolved_prompt = evolved_prompt.split("<prompt>")[1].split("</prompt>")[0]
+ return evolved_prompt
+
+
@log_calls("Updating prompts")
def update(prompts: list[str], scores: list[float], N: int):
# EVOPROMPT iteratively generates new candidate prompts and assesses each prompt
@@ -195,7 +217,7 @@ def update(prompts: list[str], scores: list[float], N: int):
return retained_prompts, retained_scores
-def run_episode():
+def run_episode(evo_alg_str: str):
# Algorithm 1 Discrete prompt optimization: EVOPROMPT
# Require:
@@ -225,23 +247,51 @@ def run_episode():
for t in trange(1, T + 1, desc="T", leave=True):
# Line 3: Selection: select a certain number of prompts from current population as parent prompts
# pr1,...,prk ∼ Pt−1
- for _ in trange(N, desc="N", leave=False):
+ if evo_alg_str == "de":
+ # DE needs best prompt for evolution
+ best_prompt_current_evolution = max(range(N), key=lambda i: S[t - 1][i])
+ for i in trange(N, desc="N", leave=False):
+ new_evolutions = []
+ new_evolutions_scores = []
+ # for both GA and DE we start with two parent prompts
pr1, pr2 = selection(P[t - 1], S[t - 1])
# Line 4: Evolution: generate a new prompt based on the selected parent prompts by leveraging LLM to perform evolutionary operators
# p′i â†Evo(pr1,...,prk)
- p_i = evolution(pr1, pr2)
- family_tree[p_i] = (pr1, pr2)
+ if evo_alg_str == "ga":
+ p_i = evolution_ga(pr1, pr2)
+ elif evo_alg_str == "de":
+ p_i = evolution_de(pr1, pr2, P[t - 1][i], best_prompt_current_evolution)
+
+ family_tree[p_i] = pr1, pr2
# Line 5: Evaluation
# s′_i ↠f(p′i,D)
s_i = f_D(p_i)
- P[t - 1].append(p_i)
- S[t - 1].append(s_i)
+ if evo_alg_str == "ga":
+ # for GA we consider all evolutions and select best afterward
+ new_evolutions.append(p_i)
+ new_evolutions_scores.append(s_i)
+ elif evo_alg_str == "de":
+ # for DE we keep the evolved prompt if it is better than the basic prompt, and use the basic prompt otherwise
+ if s_i > S[t - 1][i]:
+ new_evolutions.append(p_i)
+ new_evolutions_scores.append(s_i)
+ else:
+ new_evolutions.append(P[t - 1][i])
+ new_evolutions_scores.append(S[t - 1][i])
# Line 6: Update based on the evaluation scores
# Pt ↠{Pt−1, p′i} and St ↠{St−1, s′i}
- P_t, S_t = update(P[t - 1], S[t - 1], N)
+ if evo_alg_str == "ga":
+ # GA keeps N best prompts from current population and evolutions
+ P_t, S_t = update(
+ P[t - 1] + new_evolutions, S[t - 1] + new_evolutions_scores, N
+ )
+ elif evo_alg_str == "de":
+ # DE only keeps evolved or basic prompt in each step, whichever is better
+ P_t, S_t = new_evolutions, new_evolutions_scores
+
P.append(P_t)
S.append(S_t)
@@ -262,8 +312,6 @@ def run_episode():
family_tree = {}
if __name__ == "__main__":
options = argument_parser.parse_args()
- if options.evolution_algorithm == "de":
- raise NotImplementedError("DE is not implemented yet")
match options.evolution_engine:
case "llama2":
logger.info("Using Llama2 client as the evolution engine")
@@ -285,4 +333,4 @@ if __name__ == "__main__":
.message.content
)
- run_episode()
+ run_episode(evo_alg_str=options.evolution_algorithm)
--
GitLab